An intake system for saved content. Extracts context from reels (text or audio) via LLM processing and surfaces the purpose behind why it was saved.